Skip to contentSkip to conversation

copy_xstate_to_kernel: Fix typo which caused GDB regression

PATCH In review Kevin Buettner 4 participants 5 messages Jul 18, 2020 – Jul 21, 2020 x86

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 copy_xstate_to_kernel: Fix typo which caused GDB regression

This commit fixes a regression encountered while running the gdb.base/corefile.exp test in GDB's test suite.

In my testing, the typo prevented the sw_reserved field of struct fxregs_state from being output to the kernel XSAVES area. Thus the correct mask corresponding to XCR0 was not present in the core file for GDB to interrogate, resulting in the following behavior:

[kev@f32-1 gdb]$ ./gdb -q testsuite/outputs/gdb.base/corefile/corefile testsuite/outputs/gdb.base/corefile/corefile.core Reading symbols from testsuite/outputs/gdb.base/corefile/corefile...
[New LWP 232880]

warning: Unexpected size of section `.reg-xstate/232880' in core file.

With the typo fixed, the test works again as expected.

---
 arch/x86/kernel/fpu/xstate.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/x86/kernel/fpu/xstate.c +1 -1
index 6a54e83d5589..9cf40a7ff7ae 100644
@@ -1022,7 +1022,7 @@ int copy_xstate_to_kernel(void *kbuf, struct xregs_state *xsave, unsigned int of
 		copy_part(offsetof(struct fxregs_state, st_space), 128,
 			  &xsave->i387.st_space, &kbuf, &offset_start, &count);
 	if (header.xfeatures & XFEATURE_MASK_SSE)
-		copy_part(xstate_offsets[XFEATURE_MASK_SSE], 256,
+		copy_part(xstate_offsets[XFEATURE_SSE], 256,
 			  &xsave->i387.xmm_space, &kbuf, &offset_start, &count);
 	/*
 	 * Fill xsave->i387.sw_reserved value for ptrace frame:
Signature · 1 line
2.26.2
#1 lore raw

Just adding Linus, as Al is oft distracted.

Dave.

Quoted 39 lines

This commit fixes a regression encountered while running the gdb.base/corefile.exp test in GDB's test suite.

In my testing, the typo prevented the sw_reserved field of struct fxregs_state from being output to the kernel XSAVES area. Thus the correct mask corresponding to XCR0 was not present in the core file for GDB to interrogate, resulting in the following behavior:

[kev@f32-1 gdb]$ ./gdb -q testsuite/outputs/gdb.base/corefile/corefile testsuite/outputs/gdb.base/corefile/corefile.core Reading symbols from testsuite/outputs/gdb.base/corefile/corefile...
[New LWP 232880]

warning: Unexpected size of section `.reg-xstate/232880' in core file.

With the typo fixed, the test works again as expected.

---
 arch/x86/kernel/fpu/xstate.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/x86/kernel/fpu/xstate.c +1 -2
index 6a54e83d5589..9cf40a7ff7ae 100644
@@ -1022,7 +1022,7 @@ int copy_xstate_to_kernel(void *kbuf, struct xregs_state *xsave, unsigned int of
                copy_part(offsetof(struct fxregs_state, st_space), 128,
                          &xsave->i387.st_space, &kbuf, &offset_start, &count);
        if (header.xfeatures & XFEATURE_MASK_SSE)
-               copy_part(xstate_offsets[XFEATURE_MASK_SSE], 256,
+               copy_part(xstate_offsets[XFEATURE_SSE], 256,
                          &xsave->i387.xmm_space, &kbuf, &offset_start, &count);
        /*
         * Fill xsave->i387.sw_reserved value for ptrace frame:
--

2.26.2

#2 lore raw
Quoted 1 line · Dave Airlie

Just adding Linus, as Al is oft distracted.

Already in vfs.git#fixes, actually:
commit 8d95867c8610c515ffab2913b2cb19b2c7f7f6c1

Author: Kevin Buettner <[email protected]>
Date:   Sat Jul 18 00:20:03 2020 -0700

copy_xstate_to_kernel: Fix typo which caused GDB regression

This commit fixes a regression encountered while running the gdb.base/corefile.exp test in GDB's test suite.

In my testing, the typo prevented the sw_reserved field of struct fxregs_state from being output to the kernel XSAVES area. Thus the correct mask corresponding to XCR0 was not present in the core file for GDB to interrogate, resulting in the following behavior:

[kev@f32-1 gdb]$ ./gdb -q testsuite/outputs/gdb.base/corefile/corefile testsuite/outputs/gdb.base/corefile/corefile.core Reading symbols from testsuite/outputs/gdb.base/corefile/corefile...
[New LWP 232880]

warning: Unexpected size of section `.reg-xstate/232880' in core file.

With the typo fixed, the test works again as expected.

Fixes 9e46365459331 ("copy_xstate_to_kernel(): don't leave parts of destination uninitialized")
#3 lore raw
In reply to Kevin Buettner

* Kevin Buettner:

Quoted 26 lines

This commit fixes a regression encountered while running the gdb.base/corefile.exp test in GDB's test suite.

In my testing, the typo prevented the sw_reserved field of struct fxregs_state from being output to the kernel XSAVES area. Thus the correct mask corresponding to XCR0 was not present in the core file for GDB to interrogate, resulting in the following behavior:

[kev@f32-1 gdb]$ ./gdb -q testsuite/outputs/gdb.base/corefile/corefile testsuite/outputs/gdb.base/corefile/corefile.core Reading symbols from testsuite/outputs/gdb.base/corefile/corefile...
[New LWP 232880]

warning: Unexpected size of section `.reg-xstate/232880' in core file.

With the typo fixed, the test works again as expected.

---
 arch/x86/kernel/fpu/xstate.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/x86/kernel/fpu/xstate.c +0 -0
index 6a54e83d5589..9cf40a7ff7ae 100644

arch/x86/kernel/fpu/xstate.c int copy_xstate_to_kernel(void *kbuf, struct xregs_state *xsave, unsigned int of · line 1028

Quoted 8 lines
arch/x86/kernel/fpu/xstate.c +1 -1
@@ -1022,7 +1022,7 @@ int copy_xstate_to_kernel(void *kbuf, struct xregs_state *xsave, unsigned int of
 		copy_part(offsetof(struct fxregs_state, st_space), 128,
 			  &xsave->i387.st_space, &kbuf, &offset_start, &count);
 	if (header.xfeatures & XFEATURE_MASK_SSE)
-		copy_part(xstate_offsets[XFEATURE_MASK_SSE], 256,
+		copy_part(xstate_offsets[XFEATURE_SSE], 256,
 			  &xsave->i387.xmm_space, &kbuf, &offset_start, &count);
 	/*
 	 * Fill xsave->i387.sw_reserved value for ptrace frame:

Does this read out-of-bounds, potentially disclosing kernel memory? Not if the system supports AVX, I assume.

#4 lore raw
Quoted 23 lines · On Tue, 21 Jul 2020 10:59:07 +0200 Florian Weimer <[email protected]> wrote:

* Kevin Buettner:

Quoted 21 lines

This commit fixes a regression encountered while running the gdb.base/corefile.exp test in GDB's test suite.

In my testing, the typo prevented the sw_reserved field of struct fxregs_state from being output to the kernel XSAVES area. Thus the correct mask corresponding to XCR0 was not present in the core file for GDB to interrogate, resulting in the following behavior:

[kev@f32-1 gdb]$ ./gdb -q testsuite/outputs/gdb.base/corefile/corefile testsuite/outputs/gdb.base/corefile/corefile.core Reading symbols from testsuite/outputs/gdb.base/corefile/corefile...
[New LWP 232880]

warning: Unexpected size of section `.reg-xstate/232880' in core file.

With the typo fixed, the test works again as expected.

---
 arch/x86/kernel/fpu/xstate.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
Quoted 16 lines

arch/x86/kernel/fpu/xstate.c int copy_xstate_to_kernel(void *kbuf, struct xregs_state *xsave, unsigned int of · line 1028

Quoted 13 lines
arch/x86/kernel/fpu/xstate.c +1 -1
index 6a54e83d5589..9cf40a7ff7ae 100644
@@ -1022,7 +1022,7 @@ int copy_xstate_to_kernel(void *kbuf, struct xregs_state *xsave, unsigned int of
 		copy_part(offsetof(struct fxregs_state, st_space), 128,
 			  &xsave->i387.st_space, &kbuf, &offset_start, &count);
 	if (header.xfeatures & XFEATURE_MASK_SSE)
-		copy_part(xstate_offsets[XFEATURE_MASK_SSE], 256,
+		copy_part(xstate_offsets[XFEATURE_SSE], 256,
 			  &xsave->i387.xmm_space, &kbuf, &offset_start, &count);
 	/*
 	 * Fill xsave->i387.sw_reserved value for ptrace frame:  

Does this read out-of-bounds, potentially disclosing kernel memory? Not if the system supports AVX, I assume.

An overlarge offset (first parameter) passed to copy_part() will cause fill_gap() to be called which will copy data out of &init_fpstate.xsave. Care is taken in both fill_gap and copy_part to not copy more data than the remaining count.

So, I think the answer is "no".

Kevin